For this year’s Super Bowl, The Mill in London produced 25 commercials, relying heavily on the cloud. “There’s no way we could have gotten that done without a burst of rendering in the cloud,” said The Mill group technical director Roy Trosh. “When we know we have a vendor bulge, we used to bring a [server] supplier and it took three days to get ready to render. This time it took 15 minutes.” At this week’s HPA Tech Retreat, manufacturers and users described how the industry has evolved with regard to cloud production and post.

“The use of the cloud is evolving rapidly,” said MediaSilo + Wiredrive vice president of sales Matt Thomas. “Five years ago, everyone was security conscious and concerned about the cloud. Now everyone is looking for hybrid solutions and new services are popping up.”

Sohonet chief executive Chuck Parker reported that, “even the most stringent studios are comfortable” with the cloud. “Now we’ve moved on to the challenges of pushing assets to the cloud,” he said. “We’re looking at bandwidth, sizing, caching, and other issues.” He also reported that “most of the Tier 1 visual effects companies are doing some rendering in the cloud,” with the smaller VFX companies “getting their feet wet.”

BeBop chief technology officer/co-founder Dave Benson believes we are “at a tipping point.” “People are looking at how to do that, rather than should we do it,” he said. “They’re looking for new tools to accelerate and enable it. Putting everything in the cloud may not be viable today but it will in the near future.”

Trosh reports that one of those tools his company has been experimenting with comes from Teradici, which offers a virtual machine (KVM and Wacom tablet) over IP. “But you have to be careful how far away you are, because latency is the killer,” he warned.

Cerberus technical director Chris Clarke also had to deal with latency when his company’s Live Link for production in the cloud, which was designed for use in the U.K., signed on for “Captain America: Civil War,” which was shooting in Atlanta and Berlin. Latency was an issue, and it also took months to get security clearance, said Clarke. “The studio found holes and helped us patch those holes,” he added.

“Latency is definitely the enemy,” said Benson. “Spinning up infrastructure for a purpose-built project and then turning it off when you’re done is completely viable. It’s the way you have to do things in cloud. You have to think more distributed and geographically where your workflow is. Our goal is that someone logs in and that’s all invisible.”

Thomas warned against trying to replicate on-premise services in the cloud, and called for more interoperability of the disparate cloud tools and services. Benson stressed that, regardless of where you are in the workflow, you have to think cloud. “Whether you’re planning your infrastructure or migrating workflows to the cloud, you have to think about how it works in this environment,” he said. “When you don’t, bad things happen.”